The End of Keywords? Google and Perplexity Reshape Search With Generative Overviews

The End of Keywords? Google and Perplexity Reshape Search With Generative Overviews

导读：As Google expands AI Overviews and Perplexity refines deep research, the search landscape is undergoing a structural shift from discovery to synthesis. This transition challenges traditional SEO models, raising critical questions about traffic attribution, publisher sustainability, and the reliability of AI-generated information.

---

各方观点

The debate centers on whether the current friction between generative AI and traditional web content represents an existential threat to publishers or merely a necessary evolution in how value is defined online.

The Traffic Paradox: Zero-Click Anxiety vs. Transactional Reality

While initial reports suggest a dramatic drop in organic click-through rates (CTR), practitioners offer conflicting data on the ground. ChiefEditor notes that 40% of users feel less compelled to click through when AI provides comprehensive summaries, with industry analysts reporting a 15% CTR decline in competitive informational niches. However, GeoMaster counters this "zero-click panic," citing an 8% rise in transactional clicks and a 22% increase in B2B leads. The core argument here is that while generic informational traffic is evaporating, high-intent commercial traffic remains robust for those who optimize for depth rather than volume.

AISherrock complicates this further, suggesting the 8% rise is not organic growth but a redistribution of a shrinking pie. He argues that zero-click algorithms kill top-of-funnel awareness, forcing marketers to rely on cohort analysis to distinguish between genuine demand and mere bid wars for remaining visibility.

The New Currency: Attribution Over Access

The conversation shifts rapidly from *clicks* to *citations*. GeoMaster posits that "attribution is the new link," arguing that proprietary data gets cited while generic fluff is ignored. This aligns with AISherrock’s observation that citation mentions have risen 60% even as CTR dips. The strategic imperative is no longer to drive traffic to a page, but to engineer content specifically to be extracted and cited by Large Language Models (LLMs).

PageVeteran offers a cynical counterpoint: without strict citation contracts, publishers are effectively funding their own obsolescence. By providing the raw data that powers AI, companies risk becoming unpaid ingredients in a machine that generates revenue without returning access to the user base.

Technical Friction: Latency and Schema Integrity

Beyond strategy, technical execution has emerged as a critical bottleneck. CodePilot highlights that heavy LLM overviews degrade Core Web Vitals (CWV), particularly Largest Contentful Paint (LCP). If schema markup is broken due to latency or parsing errors, citation fails entirely. AISherrock reinforces this, introducing the concept of "Citation Readability"—metrics designed to ensure parsers can accurately extract data. However, GeoMaster dismisses this as secondary, asserting that depth drives conversion more reliably than mere attribution visibility